Message passing for GPGPU clusters: CudaMPI
- 1 January 2009
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15525244,p. 1-8
Abstract
We present and analyze two new communication libraries, cudaMPI and glMPI, that provide an MPI-like message passing interface to communicate data stored on the graphics cards of a distributed-memory parallel computer. These libraries can help applications that perform general purpose computations on these networked GPU clusters. We explore how to efficiently support both point-to-point and collective communication for either contiguous or noncontiguous data on modern graphics cards. Our software design is informed by a detailed analysis of the actual performance of modern graphics hardware, for which we develop and test a simple but useful performance model.Keywords
This publication has 6 references indexed in Scilit:
- Message passing on data-parallel architectures2009
- 3D finite difference computation on GPUs using CUDA2009
- Stencil computation optimization and auto-tuning on state-of-the-art multicore architectures2008
- GPU ComputingProceedings of the IEEE, 2008
- Latency lags bandwithCommunications of the ACM, 2004
- On the design of display processorsCommunications of the ACM, 1968